Gas-Phase Reactions of Carbon Dioxide with Atomic Transition-Metal

Gas-Phase Reactions of Sulfur Hexafluoride with Transition Metal and Main Group Atomic Cations: ..... Sarah L. Broadley , Tomas Vondrak , John M. C. P...
0 downloads 0 Views 214KB Size
Download PDF
1232

J. Phys. Chem. A 2006, 110, 1232-1241

Gas-Phase Reactions of Carbon Dioxide with Atomic Transition-Metal and Main-Group Cations: Room-Temperature Kinetics and Periodicities in Reactivity† Gregory K. Koyanagi and Diethard K. Bohme* Department of Chemistry, Centre for Research in Mass Spectrometry and Centre for Research in Earth and Space Science, York UniVersity, Toronto, Ontario, Canada, M3J 1P3 ReceiVed: May 20, 2005; In Final Form: July 29, 2005

The chemistry of carbon dioxide has been surveyed systematically with 46 atomic cations at room temperature using an inductively-coupled plasma/selected-ion flow tube (ICP/SIFT) tandem mass spectrometer. The atomic cations were produced at ca. 5500 K in an ICP source and allowed to cool radiatively and to thermalize by collisions with Ar and He atoms prior to reaction downstream in a flow tube in helium buffer gas at 0.35 ( 0.01 Torr and 295 ( 2 K. Rate coefficients and products were measured for the reactions of first-row atomic ions from K+ to Se+, of second-row atomic ions from Rb+ to Te+ (excluding Tc+), and of third-row atomic ions from Cs+ to Bi+. CO2 was found to react in a bimolecular fashion by O atom transfer only with 9 early transition-metal cations: the group 3 cations Sc+, Y+, and La+, the group 4 cations Ti+, Zr+, and Hf+, the group 5 cations Nb+ and Ta+, and the group 6 cation W+. Electron spin conservation was observed to control the kinetics of O atom transfer. Addition of CO2 was observed for the remaining 37 cations. While the rate of addition was not measurable some insight was obtained into the standard free energy change, ∆G°, for CO2 ligation from equilibrium constant measurements. A periodic variation in ∆G° was observed for first row cations that is consistent with previous calculations of bond energies D0(M+-CO2). The observed trends in D0 and ∆G° are expected from the variation in electrostatic attraction between M+ and CO2 which follows the trend in atomic-ion size and the trend in repulsion between the orbitals of the atomic cations and the occupied orbitals of CO2. Higher-order CO2 cluster ions with up to four CO2 ligands also were observed for 24 of the atomic cations while MO2+ dioxide formation by sequential O atom transfer was seen only with Hf+, Nb+, Ta+, and W+.

1. Introduction Despite the global importance of carbon dioxide, surprisingly few experimental or theoretical investigations have been reported for the interactions of CO2 with atomic metal cations. This can be attributed in part to earlier difficulties in generating atomicmetal cations in the gas phase for mass-spectrometric study and to earlier shortcomings associated with computational approaches. Consequently, previous experimental studies have been restricted to individual or small groups of atomic-metal cations. Only one systematic survey has been reported for the interactions of CO2 with atomic metal cations using ab initio theory, and this was restricted to first-row transition-metal cations. In early (1971) drift-tube and flowing- afterglow massspectrometer experiments, CO2 was observed to add to alkalimetal cations Li+, Na+, and K+ produced by thermionic surfaceionization emitters.1 Apparently, the first measurement of a reaction of a transition-metal cation with CO2 in the gas phase was that with Zr+ reported in 1985 by Dheandhanoo et al.2 Zr+ ions were produced in a magnetically confined Penning discharge containing neon and ZrBr4 and which was coupled to a selected-ion drift apparatus to measure rate coefficients. The measurements were made over an energy range from thermal energies (0.04 eV) to 0.3 eV (mean kinetic energy). * Corresponding author. E-mail: [email protected]. Telephone: 416-736-2100, ext 66188. Fax: 416-736-5936. † Part of the special issue “William Hase Festschrift”.

Two sequential O atom transfer reactions, reactions 1 and 2, were observed to occur.

Zr+ + CO2 f ZrO+ + CO

(1)

ZrO+ + CO2 f ZrO2+ + CO

(2)

Reaction 1 was reported to react with k ) 4.0 × 10-10 cm3 molecule-1 s-1 at 0.04 eV decreasing to 1 × 10-10 cm3 molecule-1 s-1 at 0.3 eV. Reaction 2 was measured to proceed with k ) 1 × 10-12 cm3 molecule-1 s-1 at thermal energies, also decreasing with increasing ion energy. The first metal-cation study using ion cyclotron resonance (ICR) mass spectroscopy, in which the metal cations were produced by laser volatilization/ionization, appears to have been published in 1991 by Irikura and Beauchamp who reported the following reaction sequence for tungsten cations with rate coefficients k3 ) 6 × 10-11 cm3 molecule-1 s-1 and k4 ) 2 × 10-11 cm3 molecule-1 s-1, both estimated to be accurate to ( 25%.3

W+ + CO2 f WO+ + CO

(3)

WO+ + CO2 f WO2+ + CO

(4)

Using similar techniques, Wesendrup and Schwarz4 were able to show that, of the third-row metal ions Ta+, W+, Os+, Ir+, and Pt+, only Ta+ and W+ form oxide cations in reactions with CO2. These authors reported that reactions 5 and 6 occur at the

10.1021/jp0526602 CCC: $33.50 © 2006 American Chemical Society Published on Web 09/02/2005

Gas-Phase Reactions of Carbon Dioxide

J. Phys. Chem. A, Vol. 110, No. 4, 2006 1233

collision rate, although no measured rate coefficients were reported. +

+

Ta + CO2 f TaO + CO

(5)

TaO+ + CO2 f TaO2+ + CO

(6)

Guided-ion beam mass spectrometry has been used to investigate the activation of CO2 by U+,5a Al+,5b V+,5c Fen+ (n ) 1-18),5d Nb+,5e Crn+ (n ) 1-18),5f Mo+,5g Y+,5h Cu+,5i Zr+,5j and Pt+.5k,l High-energy thresholds were reported for the onsets of MO+ and MCO+ production in the endothermic reactions of Al+, Fe+, Cr+, Mo+, Cu+, and Pt+ with CO2. The reactions of U+, Nb+, Y+, and Zr+ were reported to proceed in an exothermic fashion by O atom transfer at, or close to, every collision near 0 eV, CM. A sharp minimum was observed in the cross section vs ion energy profile for VO+ production with a cross section of 8 × 10-18 cm2 at near 0 eV, CM. Also investigated was the activation of CO2 by the monoxide cations UO+,5a NbO+,5e MoO+,5g YO+,5h ZrO+,5f and PtO+.5k,j Of these, only the reactions of NbO+ and MoO+ exhibit exothermic reaction behavior. The clustering of carbon dioxide to FeO+ according to reaction 7 has been observed with a selected-ion flow tube (SIFT) tandem mass spectrometer, but no reaction was seen between Fe+ and CO2, k 159 kcal mol-1 which is much larger than OA(CO) ) 127.2 ( 0.1 kcal mol-1.21 The remaining 37 atomic cations all were observed to react slowly by CO2 addition and, with the exceptions of V+ and As+, all have O atom affinities less than 127.2 kcal mol-1, and this makes O atom transfer endothermic. Figure 3 shows how the reaction efficiency for O atom transfer from CO2 varies with the O atom affinity for the 46 atomic cations investigated. There was no indication of VO+ formation from V+ or AsO+ formation with As+. The absence of exothermic O atom transfer from CO2 to groundstate V+ (X5D) can be understood since this transfer reaction to form ground-state VO+ (X3Σ) is spin forbidden, as has been noted previously by Sievers et al.5a A similar situation applies to the reaction with As+ (X3P) to form AsO+ (X1Σ)22 for which our experiments also indicate exclusive formation of the CO2 adduct. Apparently no spin-allowed surfaces are accessible to either As+ (X3P) or V+ (X5D) that lead to exothermic O atom transfer. Our failure to observe O atom transfer to V+ (X5D) is somewhat at odds with the sharp minimum observed in guidedion beam experiments in the cross section vs ion energy profile for VO+ production with a cross section of 8 × 10-18 cm2 at near 0 eV, CM.5c However, the reported interpretation of this

Gas-Phase Reactions of Carbon Dioxide

J. Phys. Chem. A, Vol. 110, No. 4, 2006 1235

Figure 1. Composite of ICP/SIFT results for the reactions of the early first-row transition-metal ions Sc+, Ti+, Co+, and As+ with N2O in helium buffer gas at 0.35(0.01 Torr and 295(2 K.

latter value was somewhat qualified: “the cross section shown is due exclusively to reaction of V+(a5D) with CO2 (presumably the authors mean X5D), although O2 and V+ excited-state contamination cannot be ruled out with 100% confidence”.5c 3.4. Comparison with O Atom Abstraction from N2O. A comparison of O atom affinities shows that OA(CO) is 87 kcal mol-1 larger than OA(N2) so that endothermicity is much more constraining to O atom transfer from CO2 than from N2O. Indeed, many more O atom transfer reactions are exothermic between N2O and the 46 metal cations that were investigated. The nine O atom transfer reactions observed with CO2 also were

all observed with N2O,12 but interesting differences are present in the measured rate coefficients and reaction efficiencies. The two O atom transfer reactions with the first-row metal cations observed with both N2O and CO2 are much more efficient with N2O than with CO2: 0.73 vs 0.087 and 0.47 vs 0.049 for Sc+ and Ti+, respectively (Ti+ also reacts with N2O by N atom transfer but this has been factored out in this comparison).12 Ground-state O atom transfer to both of these metal cations is spin forbidden from both N2O and CO2, but transfers from N2O are more exothermic by 87 kcal mol-1. So we may attribute the relatively low efficiencies observed for

1236 J. Phys. Chem. A, Vol. 110, No. 4, 2006

Koyanagi and Bohme

TABLE 1: Rate Coefficients (in units of cm3 molecule-1 s-1), Reaction Efficiencies (k/kc), and Higher-order Product Ions Measured for Reactions of Atomic Cations with Carbon Dioxide in Helium at 0.35 ( 0.01 Torr and 295 ( 2 K reaction

ka

kcb

k/kc

K+ + CO2 f K+‚CO2 Ca+ + CO2 f Ca+‚CO2 Sc+ + CO2 f ScO+ + CO Ti+ + CO2 f TiO+ + CO V+ + CO2 f V+‚CO2 Cr+ + CO2 f Cr+‚CO2 Mn+ + CO2f Mn+‚CO2 Fe+ + CO2f Fe+‚CO2 Co+ + CO2f Co+‚CO2 Ni+ + CO2f Ni+‚CO2 Cu+ + CO2f Cu+‚CO2 Zn+ + CO2 f Zn+‚CO2 Ga+ + CO2 f Ga+‚CO2 Ge++ CO2f Ge+‚CO2 As++ CO2 f As+‚CO2 Se++ CO2 f Se+‚CO2 Rb+ + CO2 f Rb+‚CO2 Sr+ + CO2 f Sr+‚CO2 Y+ + CO2 f YO+ + CO Zr+ + CO2 f ZrO+ + CO Nb++ CO2 f NbO+ + CO Mo+ + CO2 f Mo+‚CO2 Ru+ + CO2 f Ru+‚CO2 Rh+ + CO2 f Rh+‚CO2 Pd+ + CO2 f Pd+‚CO2 Ag+ + CO2 f Ag+‚CO2 Cd+ + CO2 f Cd+‚CO2 In+ + CO2 f In+‚CO2 Sn+ + CO2 f Sn+‚CO2 Sb+ + CO2 f Sb+‚CO2 Te+ + CO2 f Te+‚CO2 Cs+ + CO2 f Cs+‚CO2 Ba+ + CO2 f Ba+‚CO2 La++ CO2 f LaO+ + CO Hf++ CO2 f HfO+ + CO Ta++ CO2 f TaO+ + CO W++ CO2 f WO+ + CO Re++ CO2 f Re+‚CO2 Os++ CO2 f Os+‚CO2 Ir++ CO2 f Ir+‚CO2 Pt++ CO2 f Pt+‚CO2 Au++ CO2 f Au+‚CO2 Hg++ CO2 f Hg+‚CO2 Tl+ + CO2 f Tl+‚CO2 Pb+ + CO2 f Pb+‚CO2 Bi+ + CO2 f Bi+‚CO2

g9 × 10-13 g5 × 10-13 7.4 × 10-11 4.1 × 10-11 g5 × 10-13 g6 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g4 × 10-13 g7 × 10-13 g5 × 10-13 g9 × 10-13 g1 × 10-12 g5 × 10-13 g5 × 10-13 g6 × 10-13 5.9 × 10-10 2.5 × 10-10 1.8 × 10-10 g4 × 10-13 g4 × 10-13 g4 × 10-13 g4 × 10-13 g5 × 10-13 g6 × 10-13 g5 × 10-13 g6 × 10-13 g8 × 10-13 g6 × 10-13 g4 × 10-13 g6 × 10-13 4.2 × 10-10 2.5 × 10-10 2.4 × 10-10 4.2 × 10-11 g4 × 10-13 g6 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g5 × 10-13 g4 × 10-13

8.8 × 10-10 8.5 × 10-10 8.5 × 10-10 8.3 × 10-10 8.2 × 10-10 8.2 × 10-10 8.1 × 10-10 8.1 × 10-10 8.0 × 10-10 8.0 × 10-10 8.0 × 10-10 7.7 × 10-10 7.7 × 10-10 7.6 × 10-10 7.6 × 10-10 7.5 × 10-10 7.4 × 10-10 7.4 × 10-10 7.4 × 10-10 7.4 × 10-10 7.3 × 10-10 7.3 × 10-10 7.2 × 10-10 7.2 × 10-10 7.2 × 10-10 7.2 × 10-10 7.1 × 10-10 7.1 × 10-10 7.0 × 10-10 7.0 × 10-10 7.0 × 10-10 7.0 × 10-10 6.9 × 10-10 6.9 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.7 × 10-10 6.6 × 10-10 6.6 × 10-10 6.6 × 10-10

g1 × 10-3 g6 × 10-4 0.087 0.049 g6 × 10-4 g7 × 10-4 g6 × 10-4 g6 × 10-4 g6 × 10-4 g6 × 10-4 g5 × 10-4 g9 × 10-4 g6 × 10-4 g1 × 10-3 g1 × 10-3 g7 × 10-4 g7 × 10-4 g8 × 10-4 0.80 0.34 0.25 g5 × 10-4 g5 × 10-4 g5 × 10-4 g5 × 10-4 g7 × 10-4 g8 × 10-4 g7 × 10-4 g8 × 10-4 g1 × 10-3 g8 × 10-4 g6 × 10-4 g9 × 10-4 0.61 0.37 0.36 0.063 g6 × 10-4 g9 × 10-4 g7 × 10-4 g7 × 10-4 g7 × 10-4 g7 × 10-4 g7 × 10-4 g7 × 10-4 g7 × 10-4

higher-order product ions

ScO+(CO2)1,2 TiO+‚CO2 V+(CO2)2-4 Cr+(CO2)2,3 Mn+(CO2)2 Fe+(CO2)2 Co+(CO2)2 Ni+(CO2)2,3 Cu+(CO2)2 Zn+(CO2)2 Ge+(CO2)2 As+(CO2)2 Sr+(CO2)2 YO+‚CO2 ZrO+(CO2)1,2 NbO+CO2, NbO2+(CO2)0-3 Mo+(CO2)2, Ru+(CO2)2 Rh+(CO2)2 Pd+(CO2)2 Ag+(CO2)2

Ba+(CO2)2,3 LaO+‚CO2 HfO+(CO2)1,2 TaO2+(CO2)0-4 WO2+(CO2)0-3 Re+(CO2)2 Os+(CO2)2,3 Ir+(CO2)2,3 Pt+(CO2)2+ Au+(CO2)2+ Pb+(CO2)2+ Bi+(CO2)2+

a Effective bimolecular rate coefficient first-order in CO2. The estimated uncertainty is (30%. b Collision rate coefficient calculated using the algorithm of the modified variational transition-state/ classical trajectory theory developed by Su and Chesnavich.20a

the reactions with CO2 to the greater availability of excited spinallowed products in the reactions with N2O. For the second row, the relative values for O atom transfer from N2O and CO2 are as follows: 0.82 vs 0.80, 0.54 vs 0.34, and 0.54 vs 0.25 with Y+, Zr+, and Nb+, respectively (with N atom transfer factored out). The efficiencies for the N2O and CO2 reactions are comparable, but the latter reactions, which are less exothermic by 87 kcal mol-1, are up to two times less efficient. The latter two ground-state reactions with Zr+ and Nb+ are spin forbidden so that we can expect details of the potential-energy surfaces and crossings between them, for the most part unknown, to be rate determining. The relative values for O atom transfer from N2O and CO2, respectively, with third-row atomic cations are as follows: 0.73 vs 0.61, 0.88 vs 0.37, 0.52 vs 0.36, and 0.85 vs 0.063 with La+, Hf+, Ta+, and W+, respectively (with N atom transfer factored out). The divergence in these efficiencies is substantial for the reactions with Hf+ and more so for the reactions with

W+. Of these eight reactions only the two with Hf+ are spin allowed. O atom transfer with the ground states of La+, Ta+, and W+ is spin-forbidden to from ground-state products but may become spin-allowed for the formation of excited product MO+ ions if sufficiently exothermic. This is not unlikely for the substantial exothermicities of these reactions, but is least likely with W+ since this reaction is the least exothermic (by 39 ( 10 kcal mol-1). As we have noted already, all four of the ground-state O atom transfers from N2O and CO2 to V+ and As+ are exothermic and spin forbidden but seen only with N2O at efficiencies of 0.27 and 0.83, respectively. This diverse behavior with these two isoelectronic molecules again can be understood since the two reactions of N2O are more exothermic by 87 kcal mol-1 and so are likely to form spin-allowed excited metal-monoxide ion products.12 3.5. Dioxide Formation from MO+. A second O atom abstraction, reaction 9, was observed with NbO+, TaO+, and

Gas-Phase Reactions of Carbon Dioxide

J. Phys. Chem. A, Vol. 110, No. 4, 2006 1237

TABLE 2: O Atom Affinities, D0(M+-O) in kcal mol-1, and Ionization Energies, IE(M) in eV, for the Metal Ions Taken, with Few Exceptions, from the Review by Schro1 der et al.23 first row M+ K+

Ca+ Sc+ Ti+ V+ Cr+ Mn+ Fe+ Co+ Ni+ Cu+ Zn+ Ga+ Ge+ As+ Se+ a

j

second row

OA(M+) 3 77.2 164.6 ( 1.4a 158.6 ( 1.6a 134.9 ( 3.5a 85.8 ( 2.8a 68.0 ( 3.0a 80.0 ( 1.4a 74.9 ( 1.2a 63.2 ( 1.2a 37.4 ( 3.5a 38.5 ( 1.2a 5.6 81.8 147 ( 2k 92k b

c

IE(M)

M+

4.34 6.11 6.56 6.83 6.75 6.77 7.43 7.90 7.88 7.64 7.72 9.39 6.00 7.90 9.81 9.75

Rb+

Ru+ Rh+ Pd+ Ag+ Cd+ In+ Sn+ Sb+ Te+

Sr+ Y+ Zr+ Nb+ Mo+

OA(M+)

third row IE(M)

M+

OA(M+)

IE(M)

7 71.4 167.0 ( 4.2b 178.9 ( 2.5b 164.4 ( 2.5b 116.7 ( 0.5b

4.18 5.70 6.22 6.63 6.76 7.09

Cs+

87.9 ( 1.2c 69.6 ( 1.4c 33.7 ( 2.5c 28.4 ( 1.2c

7.36 7.46 8.34 7.58 8.99 5.79 7.34 8.64 9.01

14 92.8 206 ( 4d 173 ( 5e 188 ( 15e 166 ( 10f 115 ( 15g 100 ( 12h 59e 77,i 75 ( 2j

3.89 5.21 5.58 6.83 7.55 7.86 7.83 8.44 8.97 8.96 9.23 10.44 6.11 7.42 7.29

75.1 96.6

Ba+ La+ Hf+ Ta+ W+ Re+ Os+ Ir+ Pt+ Au+ Hg+ Tl+ Pb+ Bi+

53.2 41.6

d

Reference 24. Reference 25. Reference 26. Reference 27. e Reference 21. f Reference 28. g Reference 29. h Reference 30. i Reference 31. Reference 32. k Reference 22.

Figure 2. Periodic variations observed in the efficiencies, k/kc (represented as solid circles), for reactions of atomic cations with carbon dioxide. k represents the measured reaction rate coefficient and kc is the calculated collision rate coefficient (see Table 1). Also indicated are the observed reaction channels and the ground-state electronic configurations of the atomic cations. The reactions of Tc+ and Po+ were not investigated. The numbers in parentheses indicate the number of observed sequential reactions, either O atom transfer (green) or CO2 clustering (yellow).

WO+ with k ) 0.14 × 10-10, 2.5 × 10-10, and 0.084 × 10-10 cm3 molecule-1 s-1

MO+ + CO2 f MO2+ + CO

(9)

respectively. HfO+ also was observed to form dioxide cations but only very slowly with k ) 5 × 10-12 cm3 molecule-1 s-1. Reaction 9 is known to be exothermic with NbO+, TaO+, and WO+ for which D(O-MO+) ) 132, 140, and 132 kcal mol-1, respectively.33 The most exothermic of these reactions is also the fastest. The O atom affinity of HfO+ is not known. Sequential O atom transfer with CO2 was observed to stop with the formation of the dioxide cation; higher oxide cations were not observed presumably because their formation is endothermic.

Comparisons can be made with previous kinetic results reported by others for reaction 9 with M ) Y, Zr, Nb, Mo,Ta, and W and there is reasonable agreement. Our value of (2.5 ( 0.8) × 10-10 cm3 molecule-1 s-1 (k/kc ) 0.35) for Ta+ is consistent with the previous report of reaction 9 with M ) Ta proceeding at the collision rate in an ICR-MS.4 Our value of (8.4 ( 2.5) × 10-12 cm3 molecule-1 s-1 for WO+ is slightly lower than the value of (2 ( 25%) × 10-11 cm3 molecule-1 s-1 also obtained with an ICR-MS.3 We did not observe the formation of ZrO2+ according to reaction 9, k < 10-13 cm3 molecule-1 s-1, while the selected-ion drift measurements provided a value of 2.5 (+2.5, -1.25) × 10-12 cm3 molecule-1.2 The origin of this discrepancy is not known; details

1238 J. Phys. Chem. A, Vol. 110, No. 4, 2006

Koyanagi and Bohme TABLE 3: Experimental and Computed Values for D0(M+-CO2) Binding Energies for First-Row Transition-Metal Cations M+

M+-CO2

method

ref

Sc+

24.2 20.7 27.2 21.1 21.0 18.2 17.3 ( 0.9 19.5 15.1 13.5 13.2 17.0 16.1 8(2 13.0 ( 1.3 14. 3 ( 0.9 24.5 21.3 >20.0

B3LYP CCSD(T) (ANO) B3LYP CCSD(T) (ANO) B3LYP CCSD(T) (ANO) guided-ion beam MS B3LYP CCSD(T) (ANO) B3LYP CCSD(T) (ANO) B3LYP CCSD(T) (ANO) FT-ICR MS FT-ICR MS guided-ion beam MS B3LYP CCSD(T) (ANO) photodissociation spectroscopy B3LYP CCSD(T) (ANO) photodissociation spectroscopy B3LYP CCSD(T) (ANO)

11 11 11 11 11 11 5(c) 11 11 11 11 11 11 7(a) 7(b) 7(c) 11 11 10(b)

Ti+ V+ Cr+ Mn+ Fe+

Co+

Figure 3. Dependence of the reaction efficiency, k/kc, on the O atom affinity, OA, of the cation. k represents the measured reaction rate coefficient for loss of M+ and kc is the calculated collision rate coefficient (see Table 1). The dashed line represents OA(CO) ) 127.2 ( 0.1 kcal mol-1.21 Reactions on the right of the dashed line are exothermic for O atom transfer while those on the left are endothermic. Solid points represent O atom transfer and open circles represent CO2 addition. All open circles are lower limits to the reaction efficiency.

of the drift measurement were not reported, but our results are consistent with guided ion-beam measurements which indicate endothermic behavior for the reaction of ZrO+ with CO2.5j There is further agreement with previous guided ion-beam measurements for the reactions of NbO+ and MoO+ that were found to exhibit exothermic reaction behavior and the nonreaction of YO+ that was found to be endothermic.5e,g,h We have previously observed MO2+ dioxide formation with O234 and N2O.12 The comparison with O2 is interesting because of the similarities in exothermicities (OA(O) ) 119.1 ( 0.1 kcal mol-1 vs OA(CO) ) 127.2 ( 0.1 kcal mol-1). O2 was observed to transfer an O atom to NbO+, MoO+, TaO+, and WO+ compared to O atom transfer from CO2 to HfO+, NbO+, TaO+, and WO+. There are two dissimilarities in these reactivities. HfO+ adds O2 to form HfO3+.34 The failure of CO2 compared to the ability of O2 to transfer an O atom to MoO+ suggests that 119.1 kcal mol-1 < OA(MoO+) < 127.2 kcal mol-1. Understandably more extensive and higher oxide formation was observed with N2O because OA(N2) is only 40 kcal mol-1. 3.6. CO2 Clustering. It seems that CO2 addition to atomic ions has been seen previously in the gas phase only with some main-group and first-row transition-metal cations: Mg+, Al+, Si+, V+, Fe+, Co+, and Ni+.9,10 Some of these observations led to the spectroscopic measurement of M+-CO2 bond dissociation energies, and these are listed in Table 3. Also included in Table 3 are computed values for D0(M+-CO2) and a few additional experimental values obtained for D0(Fe+-CO2) by applying the bracketing technique to selected ligand-switching reactions. The M+-CO2 bond dissociation energies summarized in Table 3 are quite low and range from ca. 10 to 25 kcal mol-1. This translates to a range in standard free energies of ligation, reaction 8b, at room temperature from ca. -4 to -19 kcal mol-1 when the standard entropy change associated with the ligation is estimated at ca. -20 cal mol-1 deg-1 which is typical, within (2 cal mol-1 deg-1, for the electrostatic ligation of small

Ni+

27.3 24.2 24.9 ( 0.2

Cu+

24.3 21.4

11 11 10(d) 11 11

molecules to atomic cations.35 This corresponds to a range in equilibrium constants for the ligation reaction from ca. 103 to 1014 (standard state ) 1 atm) at room temperature since K ) exp(-∆G°/RT). For the specific case of the CO2 ligation of Fe+ for which we can take a value for D0 of 14 kcal mol-1,7c for example, we estimate a value for ∆G°298 ) -8 kcal mol-1 which corresponds to an equilibrium constant at room temperature, K ) 6 × 105. Equilibrium constants with this approximate magnitude are measurable under the conditions of the flow tube experiments; values as large as 107 (∆G°298 ) -10 kcal mol-1) can be measured in favorable cases. The determination of an equilibrium constant for ligation reactions of type 8b is achieved at the fixed reaction time (9.4 ms) of our experiment with a plot of the ion signal ratio or ion concentration ratio, [(M+CO2)]/[(M+)], against the flow of CO2 when this plot becomes linear and mass discrimination is taken into account. Only lower limits are accessible when equilibrium is not achieved in the reaction region either because of insufficient time or because a fast sequential addition to form a higher CO2 cluster which prevents sufficient back reaction. Breakup of the adduct ion in the sampling process also will lead to lower limits for the true equilibrium constant. However, this latter effect will be most pronounced for the weakest bonds and should not be an issue for adduct ions with binding energies g5 kcal mol-1 at the short time the ions spend in the sampling region. Experiments with weak adduct ions and low rates of ligation required the addition of very large flows of CO2. At these very high flows (g1 × 1019 molecules s-1) at which CO2 becomes a significant proportion (g0.3%) of the buffer gas, the signal of the atomic ions appeared to increase and negative curvature appeared in the equilibrium ratio plots. An illustrative ratio plot is shown in Figure 4. A number of effects come into play at such high flows of CO2. Loss of ions to the wall by axial diffusion is reduced because of the presence of the heavier CO2 molecules and perhaps more so for CO2 cluster ions if they can transfer their atomic metal cation to the ambient CO2 at long range. Also, the heavier CO2 molecules may enhance collisional dissociation in the sampling region and the rate of

Gas-Phase Reactions of Carbon Dioxide

J. Phys. Chem. A, Vol. 110, No. 4, 2006 1239 TABLE 4: Standard Free Energy Changes (∆G° in kcal mol-1) and Measured Equilibrium Constants (K) for Sequential CO2 Addition Reactions to M+, MO+, and MO2+ That Exhibit Equilibrium-like Kinetics under the Operating Conditions of the SIFT Experiments (T ) 295 ( 2 K) and Where Values Are Listed as ∆G° (K)a

Figure 4. Ratio plots that illustrate the achievement of equilibrium. The curvature at high flows in the top figure is due to an observed rise in the Mo+ signal which has been attributed to diffusion. The initial sharp curvature at low flows in the bottom figure is due to rapid occurrence of the second addition of CO2.

M+

M+

the back reaction producing and so enhance the signal while reducing the cluster-ion signal and so introducing the negative curvature in the equilibrium ratio plot. As a consequence, equilibrium constants determined from linear plots that show negative curvature at high flows of CO2 are treated as lower limits. The results of the equilibrium constant measurements are listed in Table 4. Higher-order addition of CO2 to atomic cations to form M+(CO2)2 according to reaction 10 was observed in our experiments with the first-row cations M+ ) V+, Cr+, Mn+, Fe+, Co+, Ni+, Cu+, and Zn+, the second-row cations M+ ) Sr+, Mo+, Ru+, Rh+, Pd+, and Ag+, and the third-row cations M+ ) Ba+, Re+, Os+, Ir+, Pt+, Au+, Pb+, and Bi+. Formation of M+(CO2)3 according to reaction 11 was observed only with M+ ) V+, Cr+, Ni+, Ba+, Os+, and Ir+.

M+(CO2) + CO2 f M+(CO2)2 +

+

M (CO2)2 + CO2 f M (CO2)3

(10) (11)

V+(CO2)4 was the only fourth adduct that was observed. Table 1 includes a listing of all the higher-order CO2 cluster ions that were observed and Table 4 includes equilibrium data for the formation of some diadducts and triadducts of CO2. The higherorder chemistry initiated by Sc+, Fe+, Y+, and W+ can be tracked in Figure 1. Higher-order CO2 clusters of atomic metal ions have been produced previously by laser vaporization in a pulsed nozzle cluster source. These include Fe+(CO2)n with n ) 1-14,9a V+(CO2)n with n ) 1-11,9e and Ni+(CO2)n with n ) 1-14.9f,g It is interesting to note that four molecules of CO2 were determined by infrared photodissociation spectroscopy to comprise the first solvation shell of both Ni+ and V+. This suggests that the four CO2 molecules that were observed to add to V+ under our experimental conditions have completed the first solvation shell in V+(CO2)4.

M+

0-1

K+ Ca+ V+ Cr+ Mn+ Fe+ Co+ Ni+ Cu+ Zn+ Ga+ Ge+ As+ Sr+ Mo+ Ru+ Rh+ Pd+ Ag+ Cd+ In+ Sn+ Sb+ Cs+ Ba+ Re+ Pt+ Au+ Hg+ Tl+ Pb+ Bi+

-5.0 (4.4 × 103) -5.6 (1.4 × 104) -7.7 (4.1 × 105) -6.4 (4.9 × 104) -5.7 (1.4 × 104) e-6.1 (g3.0 × 104) e-7.3 (g2.3 × 105) e-6.8 (g9.9 × 104) e-6.7 (g8.1 × 104) -6.0 (2.7 × 104) -4.9 (4.1 × 103) -6.2 (3.8 × 104) -7.9 (6.5 × 105) -4.8 (3.2 × 103) e-6.0 (g2.6 × 104) e-6.0 (g2.6 × 104) e-6.0 (g2.6 × 104) e-5.9 (g2.1 × 104) e-5.8 (g1.9 × 104) -5.1 (5.4 × 103) e-3.9 (g7.7 × 102) -4.9 (3.7 × 103) -5.3 (8.2 × 103) e-1.6 (g1.4 × 101) e-5.2 (g6.2 × 103) -4.8 (3.4 × 103) e-6.4 (g5.2 × 104) e-6.2 (g3.7 × 104) -5.1 (5.4 × 103) -3.1 (1.9 × 102) e-3.6 (g4.3 × 102) e-2.6 (g8.1 × 101)

MO+ +

ScO YO+ ZrO+ NbO+ LaO + HfO+ OsO+ MO2+

1-2

-8.5 (1.6 × 106) -7.8 (5.1 × 105)

-6.9 (1.1 × 105)

-9.0 (3.8 × 106) -9.3 (6.3 × 106) -8.9 (3.2 × 106) -8.6 (2.2 × 106)

-4.8 (3.1 × 103) -7.5 (3.1 × 105) -7.5 (3.4 × 105) -7.5 (3.2 × 105) -7.5 (2.9 × 105) -6.7 (8.3 × 104)

e-6.3 (g4.2 × 104) -4.5 (1.9 × 103) e-8.9 (g3.1 × 106) e-8.1 (g9.4 × 105)

0-1

1-2

-7.5 (2.9 × ) -10.1 (2.5 × 107) -7.7 (4.4 × 105) -6.4 (5.2 × 104) -6.1 (3.0 × 104) e-7.4 (g2.5 × 105) e-9.2 (g6.1 × 106)

-8.3 (1.2 × 106)

105

0-1

2-3

-8.9 (3.7 × 106) e-7.2 (g2.0 × 105) 1-2

2-3

NbO2+ e-8.7 (g2.2 × 106) e-9.8 (g1.4 × 107) TaO2+ -8.9 (3.2 × 106) -10.2 (3.2 × 107) -8.6 (2.1 × 106) WO2+ e-8.6 (g2.2 × 106) e-9.1 (g4.5 × 106) a The estimated uncertainty in ∆G° is less than 0.3 kcal mol-1. The estimated uncertainty in K is up to (30%. Standard state ) 1 atm.

Also included in Table 4 are equilibrium data obtained for the CO2 clustering of some metal-oxide and metal-dioxide cations that was observed to occur according to reactions 12 and 13.

MO+(CO2)n + CO2 f MO+(CO2)n+1

(12)

MO2+(CO2)n + CO2 f MO2+(CO2)n+1

(13)

ZrO+(CO2) has been synthesized previously in a molecular beam and spectroscopically characterized,10e as has VO+(CO2)n+ produced in a pulsed nozzle cluster source with n ) 1-10.9e Furthermore, the clusters FeO+(CO2)1,2 have been observed to be formed in reactions of type 5 with a selected-ion flow tube (SIFT) tandem mass spectrometer at 294 K.6 The observation

1240 J. Phys. Chem. A, Vol. 110, No. 4, 2006

Koyanagi and Bohme

Figure 5. Variation in the standard free energy change, ∆G°, for the addition of CO2 to atomic-metal cations across the periodic table. Values for ∆G° are based on equilibrium constants measured under the operating conditions of the SIFT experiments (T ) 295 ( 2 K). Also indicated are bond dissociation energies (open circles, top right-hand scale) computed at the CCSD(T) (ANO) level of theory for first-row transition-metal cations.11

of MO2+(CO2)n cluster ions also has been reported, but these were produced in mixtures of CO2 and O2.9f In the case of ZrO+(CO2) and NbO+(CO2) we can compare the reported values5e,j for D0(OZr+-CO2) ) 0.74 ( 0.06 eV or 17 ( 1 kcal mol-1 and D0(ONb+-CO2) ) 0.88 ( 0.03 eV or 20 ( 1 kcal mol-1 with our measured value for ∆G° ) 7.7 and 6.4 kcal mol-1, respectively (see Table 4). When the standard entropy changes associated with the dissociations are estimated at ca. 20 cal mol-1 deg-1, our values for ∆G° reduce to ∆H° ) 14 ( 2 and 14 ( 2 kcal mol-1, respectively, at 295 K. 3.7. Periodic Trends in the Standard Free Energy for CO2 Ligation of M+. Although the rate coefficients for the ligation of the atomic-metal cations with CO2 were generally too small to measure with any accuracy, k < e 9 × 10-13 cm3 molecule-1 s-1, equilibrium constants, and therefore standard free energies of ligation, were more accessible to measurement. The latter are generally unknown from previous measurements with the exception of the computed and experimental values for D0(M+- CO2) listed in Table 3. Figure 5 explores the periodic variation in the values of D0(M+-CO2) and ∆G° listed in Tables 3 and 4, respectively, for the addition of CO2 to M+. Some clear trends are apparent and, for the first-row transition-metal

cations, there is good correspondence between the periodic behavior of ∆G° and D0(M+-CO2). Both are not unexpected. The correspondence between the periodic behavior of ∆G° and D0(M+-CO2) is expected since ∆(∆S°) ≈ 0 for the family of similar CO2 addition reactions with transition-metal cations since ∆(∆G°) ≈ ∆(∆H°) ≈ ∆(D0(M+-CO2)). The periodic trend in the calculated values for D0(M+-CO2) for first-row transitionmetal ions is similar to the periodic trends in experimental values for D0(M+-L) reported previously for other ligands such as water, ammonia, benzene, imidazole, and adenine.36 As has been noted previously,11 cationic metals are bound to CO2 electrostatically in an end-on (η1-O) coordination since CO2 has a negative quadrupole moment. The electrostatic interaction between the atomic-metal cation and CO2 has been computed to produce only small asymmetry in the two CO bond lengths which remains almost constant across the first row.11 There is no significant π-back-donation from the metal cation to CO2. The M+-CO2 bond distances are determined both by the size of the ion, which decreases across the row, and by the metalligand repulsion, which depends on the electronic configuration of the metal cation.11 The M+-CO2 binding energies across the first row of the periodic table are similarly determined by

Gas-Phase Reactions of Carbon Dioxide the trend in electrostatic attraction which follows the trend in ion size and the trend in repulsion between the metal d orbitals and the occupied orbitals of CO2, the order of repulsion being 3dσ > 3dπ > 3dδ. Minima in D0 or ∆G° appear for Mn+(3d54s1) and Ga+(3d104s24p0) where the repulsion is largest. The increase in M+-CO2 attraction due to decreasing size determines the trend in increasing bond energy observed just beyond Mn+ and Ga+ and also beyond In+. 4. Conclusions A systematic survey of the room-temperature ground-state chemistry of 46 metal cations has shown that carbon dioxide reacts in a bimolecular fashion by O atom transfer only with 9 early transition-metal cations: the group 3 cations Sc+, Y+, and La+, the group 4 cations Ti+, Zr+, and Hf+, the group 5 cations Nb+ and Ta+, and the group 6 cation W+. The efficiencies for O atom abstraction are in the range 0.046 (for Zr+) to 0.80 (for Y+). Thermodynamics and spin play a role in controlling the kinetics and product formation in O atom transfer reactions of atomic-metal cations with CO2, just as we had found previously for O atom transfer from the isoelectronic N2O molecule to the same metal cations.11 MO2+ dioxide formation by sequential O atom transfer occurs only with Hf+, Nb+, Ta+, and W+. Addition of CO2 occurs under SIFT conditions with 37 cations, those that do not react by O atom transfer. The standard free energy change ∆G° for CO2 ligation, as determined from equilibrium constant measurements, shows a periodic variation. For first row cations this variation is consistent with previous calculations of M+-CO2 bond energies and what is expected from the variation in electrostatic attraction between M+ and CO2. The latter follows the trend in atomic-ion size and the trend in repulsion between the orbitals of the atomic cations and the occupied orbitals of CO2. Higher-order CO2 cluster ions with up to four CO2 ligands are formed under SIFT conditions with 24 of the atomic cations. Acknowledgment. Continued financial support from the Natural Sciences and Engineering Research Council of Canada is greatly appreciated. Also, we acknowledge support from the National Research Council, the Natural Science and Engineering Research Council and MDS SCIEX in the form of a Research Partnership grant. As holder of a Canada Research Chair in Physical Chemistry, D.K.B. thanks the contributions of the Canada Research Chair Program to this research. References and Notes (1) (a) Keller, G. E.; Beyer, R. A. J. Geophys. Res. 1971, 76, 289. (b) Keller, G. E.; Beyer, R. A. Trans. Am. Geophys. Union Res. 1971a, 52, 303. (c) Perry, R. A.; Rowe, B. R.; Viggiano, A. A.; Albritton, D. L.; Ferguson, E. E.; Fehsenfeld, F. C. Geophys. Res. Lett. 1980, 7, 693. (2) Dheandhanoo, S.; Chatterjee, B. K.; Johnson, R. J. Chem. Phys. 1985, 83, 3327. (3) Irikura, K. K.; Beauchamp, J. L. J. Phys. Chem. 1991, 95, 8344. (4) Wesendrup, R.; Schwarz, H. Angew. Chem., Int. Ed. Engl. 1995, 34, 2033. (5) (a) Armentrout, P. B.; Beauchamp, J. L. Chem. Phys. 1980, 50, 27. (b) Clemmer, D. E.; Weber, M. E.; Armentrout, P. B. J. Phys. Chem. 1992, 96, 10888. (c) Sievers, M. R.; Armentrout, P. B. J. Chem. Phys. 1995, 102, 754. (d) Griffin, J. B.; Armentrout, P. B. J. Chem. Phys. 1997, 107, 5345 (e) Sievers, M. R.; Armentrout, P. B. Int. J. Mass Spectrom. 1998, 179/180, 103. (f) Griffin, J. B.; Armentrout, P. B. J. Chem. Phys. 1998, 108, 8075. (g) Sievers, M. R.; Armentrout, P. B. J. Phys. Chem. A. 1998, 102, 10754. (h) Sievers, M. R.; Armentrout, P. B. Inorg. Chem. 1999, 38, 397. (i) Rodgers, M. T.; Walker, B.; Armentrout, P. B. Int. J. Mass Spectrom. 1999, 182/183, 99. (j) Sievers, M. R.; Armentrout, P. B. Int. J. Mass Spectrom. 1999, 185/186/187, 117. (k) Zhang, X.-G.; Armentrout, P. B.

J. Phys. Chem. A, Vol. 110, No. 4, 2006 1241 J. Phys. Chem. A 2003, 107, 8904. (l) Zhang, X.-G.; Armentrout, P. B. J. Phys. Chem. A 2003, 107, 8915. (6) Baranov, V.; Javahery, G.; Hopkinson, A. C.; Bohme, D. K. J. Am. Chem. Soc. 1995, 117, 12801. (7) (a) Schwarz, J.; Schwarz, H. Organometallics 1994, 13, 1518. (b) Dieterle, M.; Harvey, J. N.; Heinemann, C.; Schwarz, J.; Schrder, D.; Schwarz, H. Chem. Phys. Lett. 1997, 277, 399. (c) Tjelta, B. L.; Walter, D.; Armentrout, P. B. Int. J. Mass Spectrom. 2001, 204, 7. (8) (a) Weinheimer, C. J.; Lisy, J. M. J. Phys. Chem. 1996, 100, 15305. (b) Weinheimer, C. J.; Lisy, J. M. Chem Phys. 1998, 239, 357. (9) (a) Gregoire, G.; Duncan, M. A. J. Chem. Phys. 2002, 117, 2120. (b) Gregoire, G.; Brinkmann, N. R.; van Heijnsbergen, Schaefer, H. F.; Duncan, M. A. J. Phys. Chem. A 2003, 107, 218. (c) Walters. R. S.; Brinkmann, N. R.; Schaefer, H. F.; Duncan, M. A. J. Phys. Chem. A 2003, 107, 7396. (d) Jaeger, J. B.; Jaeger, T. D.; Brinkmann, N. R.; Schaefer, H. F.; Duncan, M. A. Can. J. Chem. 2004, 82, 934. (e) Walker, N. R.; Walters. R. S.; Duncan, M. A. J. Chem. Phys. 2004, 120, 10037. (f) Walker, N. R.; Grieves, G. A.; Walters, R. S.; Duncan, M. A. Chem. Phys. Lett. 2003, 380, 230. (g) Walker, N. R.; Walters. R. S.; Grieves, G. A.; Duncan, M. A. J. Chem. Phys. 2004, 121, 10498. (10) (a) Lessen, D. E.; Asher, R. L.; Brucat, P. J. J. Chem. Phys. 1991, 95, 1414. (b) Asher, R. L.; Bellert, D.; Buthelezi, T.; Brucat, P. J. Chem. Phys. Lett. 1994, 227, 623. (c) Bellert, D.; Buthelezi, T.; Brucat, P. J. Chem. Phys. Lett. 1998, 290, 316. (d) Asher, R. L.; Bellert, D.; Buthelezi, T.; Weerasekera, G.; Brucat, P. J. Chem. Phys. Lett. 1994, 228, 390. (e) Bellert, D.; Buthelezi, T.; Hayes, T.; Brucat, P. J. Chem. Phys. Lett. 1997, 276, 242. (11) Sodupe, M.; Branchadell, V.; Rosi, M.; Bauschlicher, C. W. J. Phys. Chem. 1997, 101, 7854. (12) Lavrov, V. V.; Blagojevic, V.; Koyanagi, G. K.; Orlova, G.; Bohme, D. K. J. Phys. Chem. A 2004, 108, 5610. (13) (a) Koyanagi, G. K.; Lavrov, V.; Baranov, V. I.; Bandura, D.; Tanner, S. D.; McLaren, J. W.; Bohme, D. K. Int. J. Mass Spectrom. 2000, 194, L1. (b) Koyanagi, G. K.; Baranov, V. I.; Tanner, Bohme, D. K. J. Anal. At. Spectrom. 2000, 15, 1207. (14) Moore, C. E. Atomic energy leVels as deriVed from the analyses of optical spectra; U.S. National Bureau of Standards: Washington, DC, 1971. (15) Van Kleef, Th. A. M.; Metsch, B. C. Physica 1978, C 95, 251. (16) Lavrov, V. V.; Blagojevic, V.; Koyanagi, G. K.; Bohme, D. K. J. Phys. Chem. A 2004, 108, 5610. (17) Condon, E. U.; Shortley, G. H. The theory of atomic spectra; Cambridge University Press: Cambridge, U.K., 1963; pp 236-237. (18) Garstang, R. H. Mon. Not. R. Astron. Soc. 1962, 124, 321. (19) (a) Mackay, G. I.; Vlachos, G. D.; Bohme, D. K.; Schiff, H. I. Int. J. Mass Spectrom. Ion Phys. 1980, 36, 259. (b) Raksit, A. B.; Bohme, D. K. Int. J. Mass Spectrom. Ion Processes 1983/84, 55, 69. (20) (a) Su, T.; Chesnavich, W. J. J. Chem. Phys. 1982, 76, 5183. (b) Handbook of Chemistry and Physics, 78th ed.; CRC Press: Boca Raton, FL, 1997. (21) Lias, S. G.; Bartmess, J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. G. J. Phys. Chem. Ref. Data 1988, 17, Suppl 1. (22) S. Petrie, private communications. The calculations were done using the QCISD(T)/6-311+G(3df) method; the value for OA(Se+) does not include a spin-orbit correction. (23) Schro¨der, D.; Schwarz, H.; Shaik, S. Structure and Bonding (Berlin); Springer-Verlag: Berlin, and Heidelberg, Germany, 2000; Vol 97, pp 91122. (24) Freiser, B. S., Ed. Organnometallic ion chemistry; Kluwer: Dordrecht, The Netherlands, 1996. (25) Sievers, M. R.; Chen, Y.-M.; Armentrout, P. B. J. Chem. Phys. 1996, 105, 6322. (26) Chen, Y.-M.; Armentrout, P.B. J. Chem. Phys. 1995, 103, 618. (27) Clemmer, D. E.; Dalleska, N. F.; Armentrout, P. B. Chem. Phys. Lett. 1992, 190, 259. (28) Blagojevic, V.; Lavrov, V. V.; Orlova, G.; Bohme, D. K. Chem. Phys. Lett. 2004, 389, 303. (29) Beyer, M. Dissertation, Technical University Mu¨nchen, 1996. (30) Irikura, K. K.; Beauchamp, J. L. J. Am. Chem. Soc. 1989, 111, 75. (31) Pavlov, M.; Blomberg, M. R. A.; Siegbahn, P. E. M.; Wesendrup, R.; Heinemann, C.; Schwarz, H. J. Phys. Chem. A 1997, 101, 1567. (32) Zhang, X.-G.; Armentrout, P. B. J. Phys. Chem. A 2003, 107, 8904. (33) Schro¨der, D.; Schwarz, H.; Shaik, S. Structure and Bonding (Berlin); Springer-Verlag: Berlin, and Heidelberg, Germany, 2000; Vol. 97, pp 91122. (34) Koyanagi, G. K.; Caraiman, D.; Blagojevic, V.; Bohme, D. K. J. Phys. Chem. A 2002, 106, 4581. (35) Kebarle, P. Annu. ReV. Phys. Chem. 1977, 28, 445. (36) Rodgers, M. T.; Armentrout, P. B. Acc. Chem. Res. 2004, 37, 989.